Article
Engineering, Biomedical
Kaan Aykut Kabakci, Asli Cakir, Ilknur Turkmen, Behcet Ugur Toreyin, Abdulkerim Capar
Summary: This study proposes a cell-based image analysis method to automatically determine CerbB2/HER2 scores in breast tissue images in accordance with ASCO/CAP recommendations, providing an explainable artificial intelligence solution. The results suggest that the proposed method is highly effective in HER2 tissue scoring.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2021)
Article
Pathology
Charles J. Robbins, Aileen I. Fernandez, Gang Han, Serena Wong, Malini Harigopal, Mirna Podoll, Kamaljeet Singh, Amy Ly, M. Gabriela Kuba, Hannah Wen, Mary Ann Sanders, Jane Brock, Shi Wei, Oluwole Fadare, Krisztina Hanley, Julie Jorns, Olivia L. Snir, Esther Yoon, Kim Rabe, T. Rinda Soong, Emily S. Reisenbichler, David L. Rimm
Summary: This study evaluated the concordance and interrater reliability of scoring HER2 IHC in breast cancer biopsies by 18 breast cancer-specialized pathologists. The results showed poor concordance within the intermediate categories and for the IHC 0 cases. The use of a 3-category system partially reduced the discordance. The study also revealed the optimal number of raters for determining HER2 IHC scores of 0 and 3+.
Article
Computer Science, Artificial Intelligence
Suman Tewary, Sudipta Mukhopadhyay
Summary: This work presents an automated scoring method for prognostic marker HER2 stained tissue samples. Two CNN networks and statistical decision fusion are used to score the sample image. The proposed method achieves improved accuracy compared to existing methods through experiments and evaluations.
APPLIED SOFT COMPUTING
(2022)
Article
Medicine, General & Internal
Christiane Palm, Catherine E. E. Connolly, Regina Masser, Barbara Padberg Sgier, Eva Karamitopoulou, Quentin Simon, Beata Bode, Marianne Tinguely
Summary: This study evaluated the use of a digital and AI-assisted workflow for determining HER2 status in breast cancer patients. The results showed that this method was in accordance with the ASCO/CAP guidelines and had high agreement between pathologists and AI.
Article
Oncology
Julia R. Naso, Tetiana Povshedna, Gang Wang, Norbert Banyi, Calum MacAulay, Diana N. Ionescu, Chen Zhou
Summary: PD-L1 expression in non-small cell lung cancer is predictive of response to immunotherapy, and automated PD-L1 scoring using QuPath software shows excellent correlation with manual scoring by pathologists. However, automated scoring tends to result in more 1-49% scores compared to manual scoring. Additionally, automated scoring shows high sensitivity but lower specificity at a 1% threshold, and excellent specificity but lower sensitivity at a 50% threshold.
PATHOLOGY & ONCOLOGY RESEARCH
(2021)
Article
Oncology
Mohammad Yosofvand, Sonia Y. Khan, Rabin Dhakal, Ali Nejat, Naima Moustaid-Moussa, Rakhshanda Layeequr Rahman, Hanna Moussa
Summary: Tumor-infiltrating lymphocytes play a critical role in cancer immunotherapy, and in this study, an automated method using a two-stage deep learning model was developed to score lymphocytes in breast cancer histopathology slides. The accuracy of the method was verified by comparing results with those from expert pathologists.
Article
Oncology
Abeer M. Mahmoud, Eileen Brister, Odile David, Klara Valyi-Nagy, Maria Sverdlov, Peter H. Gann, Sage J. Kim
Summary: We trained and validated a machine learning digital scoring method for PRMT6 protein expression in lung cancer tissues, which showed excellent concordance with manual scoring by pathologists. This optimized digital scoring can serve as a more efficient and accurate method for evaluating PRMT6 expression.
Article
Pathology
Mustafa Yousif, Yiyuan Huang, Andrew Sciallis, Celina G. Kleer, Judy Pang, Brian Smola, Kalyani Naik, David S. McClintock, Lili Zhao, Lakshmi P. Kunju, Ulysses G. J. Balis, Liron Pantanowitz
Summary: This study compared the concordance between quantitative image analysis (QIA) and pathologists' scoring in evaluating the expression of estrogen receptor (ER), progesterone receptor (PgR), and human epidermal growth factor receptor 2 (HER2) in breast cancer. The results showed that QIA had excellent concordance with pathologists' scores for ER, PgR, and HER2. Pathologist oversight of representative region selection is recommended to avoid errors.
AMERICAN JOURNAL OF CLINICAL PATHOLOGY
(2022)
Article
Oncology
Jeppe Thagaard, Elisabeth Specht Stovgaard, Line Grove Vognsen, Soren Hauberg, Anders Dahl, Thomas Ebstrup, Johan Dore, Rikke Egede Vincentz, Rikke Karlin Jepsen, Anne Roslind, Iben Kumler, Dorte Nielsen, Eva Balslev
Summary: Triple-negative breast cancer (TNBC) is a difficult-to-treat cancer type with lower survival rates, with stromal tumor-infiltrating lymphocytes (sTIL) emerging as a strong prognostic biomarker for overall survival. Automated image analysis can quantify sTIL density effectively, showing high concordance with manual scoring and potential for clinical use.
Article
Medicine, General & Internal
Mohammad Faizal Ahmad Fauzi, Wan Siti Halimatul Munirah Wan Ahmad, Mohammad Fareed Jamaluddin, Jenny Tung Hiong Lee, See Yee Khor, Lai Meng Looi, Fazly Salleh Abas, Nouar Aldahoul
Summary: This study presents an automated evaluation system for determining the hormone receptor status of breast cancer patients. The system detects and classifies cells in breast cancer images and calculates the Allred score for estrogen receptor status evaluation. The system shows promising results by providing fast and reliable assistance, potentially improving prognostic reporting standards.
Review
Oncology
Yanjun Hou, Hiroaki Nitta, Zaibo Li
Summary: HER2 intratumoral heterogeneity (ITH) is a common phenomenon in breast cancer, characterized by the coexistence of tumor cell subpopulations with different HER2 gene or protein expression. It has been associated with poor prognosis in patients receiving anti-HER2 targeted therapies and proposed as a potential mechanism for anti-HER2 resistance. HER2 ITH can be categorized into non-genetic and genetic ITH based on different HER2 genetic amplification, with genetic ITH exhibiting clustered, mosaic, and scattered distribution patterns. Digital image analysis has emerged as a promising method to accurately and objectively assess HER2 ITH.
Article
Oncology
Lan Shu, Yiwei Tong, Zhuoxuan Li, Xiaosong Chen, Kunwei Shen
Summary: HER2 1+ cases of breast cancer are similar to HER2 0 cases in terms of clinicopathological features and mRNA expression, but different from HER2 2+/FISH- cases. There is poor concordance between IHC/FISH and qRT-PCR for HER2-negative tumors. The current definition of HER2-low expression with the lower boundary of HER2 IHC 1+ may be inaccurate.
Article
Cell Biology
T. Jagomast, C. Idel, L. Klapper, P. Kuppler, L. Proppe, S. Beume, M. Falougy, D. Steller, S. G. Hakim, A. Offermann, M. C. Roesch, K. L. Bruchhage, S. Perner, J. Ribbat-Idel
Summary: Quantifying protein expression in immunohistochemically stained histological slides is crucial for oncologic research. This study compared the results obtained from manual and automated digital image analysis systems and found high correlation and agreement between the two methods in measuring chromogenic intensity and positive index. Both methods were shown to be reliable for patient evaluation and aDIA was preferred due to its time-saving and reproducibility advantages.
HISTOLOGY AND HISTOPATHOLOGY
(2022)
Article
Biochemistry & Molecular Biology
Janos Bencze, Mate Szarka, Balazs Koti, Woosung Seo, Tibor G. G. Hortobagyi, Viktor Bencs, Laszlo V. Modis, Tibor Hortobagyi
Summary: This study tested a newly established artificial intelligence (AI)-aided digital image analysis platform, Pathronus, and compared it to conventional scoring. The results showed that Pathronus provides a more accurate alternative for protein quantification.
Article
Oncology
Henrik Failmezger, Harald Hessel, Ansh Kapil, Guenter Schmidt, Nathalie Harder
Summary: Identifying new tumor biomarkers is crucial in cancer research. Our novel approaches for quantitatively scoring spatial marker expression heterogeneity outperform expression averages, showing the importance of considering spatial variability in tumor biology research.
FRONTIERS IN ONCOLOGY
(2022)
Article
Mathematics, Interdisciplinary Applications
Anibal Pedraza, Oscar Deniz, Gloria Bueno
Summary: Adversarial examples pose a significant threat to machine learning models, and small perturbations can lead to model failure. To defend against adversarial examples, this study proposes an adversarial example detection method based on chaos theory, which detects adversarial perturbations by evaluating chaotic behavior in deep networks.
CHAOS SOLITONS & FRACTALS
(2022)
Article
Computer Science, Artificial Intelligence
Anibal Pedraza, Oscar Deniz, Gloria Bueno
Summary: The phenomenon of Adversarial Examples, where deep neural networks can be fooled by imperceptible perturbations, exists in the real world without maliciously selected noise. Through comparisons using distance and image quality metrics, it was shown that natural adversarial examples have a greater distance from the originals compared to artificially generated ones.
INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS
(2022)
Article
Chemistry, Analytical
Haipeng Li, Ramakrishnan Mukundan, Shelley Boyd
Summary: Breast density is an important biomarker for indicating the risk of developing breast cancer, and accurate classification of breast density is crucial for the development of a computer-aided detection system. This study proposes a novel texture descriptor and feature vector to improve the robustness and classification accuracy of mammogram features.
Article
Microscopy
Harbinder Singh, Gabriel Cristobal, Gloria Bueno, Saul Blanco, Simrandeep Singh, P. N. Hrisheekesha, Nitin Mittal
Summary: This paper presents a microscopy image fusion approach to retrieve the complete dynamic range of diatoms with complex cell walls and patterns. The method preserves details in poorly and brightly illuminated regions of the diatom shells by selecting well-exposed regions and improving histogram equalization. The proposed fusion method outperforms state-of-the-art algorithms in quantitative and qualitative assessments.
Review
Anatomy & Morphology
Jesus Salido, Gloria Bueno, Jesus Ruiz-Santaquiteria, Gabriel Cristobal
Summary: This article reviews 23 studies conducted in the last decade that focus on providing open hardware optical microscopes as a low-cost alternative to commercial systems. These studies aim to address various needs, such as limited resource institutions' research in the field of bio sciences, disease diagnosis and health screenings in developing countries, and educational settings that require high equipment availability and low replacement cost. The analysis of the selected works classifies the solutions into two main categories: portable field microscopes and multipurpose automated microscopes. Furthermore, the article discusses the maturity level of these solutions in adopting practices aligned with the development of Open Science.
MICROSCOPY RESEARCH AND TECHNIQUE
(2022)
Review
Oncology
Mohsin Bilal, Mohammed Nimir, David Snead, Graham S. Taylor, Nasir Rajpoot
Summary: Immunotherapy, particularly in the context of colorectal cancer, has seen significant advancements and the integration of artificial intelligence (AI) to predict biomarkers has been explored. Microsatellite instability (MSI) is a popular biomarker in this field, and AI has been utilized to predict MSI status and tumor mutation burden. This survey also discusses the potential of AI in predicting immune cell infiltrates and utilizing alternate data modalities such as immunohistochemistry and gene expression. The future directions emphasize the promise of AI in accelerating exploration and benefiting patients in the field of immunotherapy for colorectal cancer.
BRITISH JOURNAL OF CANCER
(2023)
Article
Pathology
Mahmoud Ali, Harriet Evans, Peter Whitney, Fayyaz Minhas, David R. J. Snead
Summary: The archiving of whole slide images is a challenge in digital pathology implementation due to the large amount of data generated. By examining the combination of SNOMED codes, it is possible to identify which cases are likely to be recalled, thus reducing the number of archived cases.
JOURNAL OF CLINICAL PATHOLOGY
(2023)
Review
Cell Biology
Ian O. Ellis, Emad A. Rakha, Gary M. Tse, Puay Hoon Tan
Summary: The International Collaboration on Cancer Reporting (ICCR) has developed an international dataset for breast cancer pathology reporting, aiming to provide a unified approach worldwide. The dataset includes essential and optional data items based on a critical review and discussion of current evidence. The process concludes with international public consultation and publication on the ICCR website to promote high-quality, standardised pathology reporting globally.
Article
Computer Science, Artificial Intelligence
Simon Graham, Quoc Dang Vu, Mostafa Jahanifar, Shan E. Ahmed Raza, Fayyaz Minhas, David Snead, Nasir Rajpoot
Summary: The recent advances in deep learning have greatly improved the performance of image analysis for digitised pathology slides. However, these deep models are often trained for a single task and require a large amount of training data. In this paper, a multi-task learning approach is proposed to address these issues by leveraging data from multiple sources and achieving simultaneous prediction for different tasks. The learned representation is also shown to be transferable for additional tasks.
MEDICAL IMAGE ANALYSIS
(2023)
Article
Computer Science, Interdisciplinary Applications
Jesus Salido, Noelia Vallez, Lucia Gonzalez-Lopez, Oscar Deniz, Gloria Bueno
Summary: This paper compares three generative models of digital staining in the H&E modality applied to breast tissue. The results show that the Cycle GAN model outperforms the other two models in terms of structural similarity and chromatic discrepancy.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2023)
Article
Oncology
Shorouk Makhlouf, Noorul Wahab, Michael Toss, Asmaa Ibrahim, Ayat G. Lashen, Nehal M. Atallah, Suzan Ghannam, Mostafa Jahanifar, Wenqi Lu, Simon Graham, Nigel P. Mongan, Mohsin Bilal, Abhir Bhalerao, David Snead, Fayyaz Minhas, Shan E. Ahmed Raza, Nasir Rajpoot, Emad Rakha
Summary: In this study, artificial intelligence (AI) was used to assess the prognostic significance of tumour infiltrating lymphocytes (TILs) in a large cohort of luminal breast cancer patients. The results showed that TILs counts and their spatial distribution had predictive value for the prognosis of early-stage breast cancer patients.
BRITISH JOURNAL OF CANCER
(2023)
Article
Surgery
Christopher Corden, Radu Boitor, Palminder Kaur Dusanjh, Andrew Harwood, Abhik Mukherjee, Dhanwant Gomez, Ioan Notingher
Summary: This study investigates the use of auto-fluorescence (AF) and Raman spectroscopy for ex vivo discrimination of colorectal liver metastases (CRLM) from normal liver tissue. The results demonstrate that AF imaging and Raman spectroscopy can effectively differentiate CRLM from normal liver tissue. These findings suggest the potential development of integrated multimodal AF-Raman techniques for intraoperative assessment of surgical margins.
JOURNAL OF SURGICAL RESEARCH
(2023)
Article
Chemistry, Multidisciplinary
Noelia Vallez, Jose Luis Espinosa-Aranda, Anibal Pedraza, Oscar Deniz, Gloria Bueno
Summary: Microscopy scanners and AI techniques have advanced biomedicine, but challenges arise from the variety of digital file formats used. DICOM stands out as a standard that transcends internal image formats, but its clinical use is limited. This paper presents the first web viewer system that integrates WSI DICOM images and AI models, aiming to bridge the gap and improve CAD WSI processing tasks.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Interdisciplinary Applications
Nicola Altini, Michele Rossini, Sandor Turkevi-Nagy, Francesco Pesce, Paola Pontrelli, Berardino Prencipe, Francesco Berloco, Surya Seshan, Jean-Baptiste Gibier, Anibal Pedraza Dorado, Gloria Bueno, Licia Peruzzi, Mattia Rossi, Albino Eccher, Feifei Li, Adamantios Koumpis, Oya Beyan, Jonathan Barratt, Huy Quoc Vo, Chandra Mohan, Hien Van Nguyen, Pietro Antonio Cicalese, Angela Ernst, Loreto Gesualdo, Vitoantonio Bevilacqua, Jan Ulrich Becker
Summary: The study aimed to replicate the glomerular components of the Oxford Classification for IgA nephropathy using a deep learning pipeline. The results showed excellent performance in automatic glomerular segmentation and classification, with high correlation to expert labels, meeting the requirements of the Oxford Classification.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Noelia Vallez, Jesus Ruiz-Santaquiteria, Oscar Deniz, Jeff Hughes, Scott Robertson, Kreshnik Hoti, Gloria Bueno
Summary: The identification of pain expression by facial analysis is challenging. This study proposes using CNNs pre-trained on pain detection models to combine different AUs for pain recognition. The method shows variability in AU detection.
IMAGE ANALYSIS AND PROCESSING, ICIAP 2022 WORKSHOPS, PT I
(2022)
