Kevin Thandiackal
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Kevin Thandiackal has a total of 26 co-authors. Here are the top co-authors by number of shared publications:
- Orcun Goksel
Department of Information Technology, Uppsala University, Uppsala, Sweden
5 shared publications - Pushpak Pati
IBM Research Europe, Zurich, Switzerland
3 shared publications - Boqi Chen
ETH AI Center, Zurich, Switzerland; Department of Computer Science, ETH Zurich, Switzerland
2 shared publications - Guillaume Jaume
Department of Fundamental Oncology, University of Lausanne, Lausanne, Switzerland; Department of Pathology, Mass General Brigham, Harvard Medical School, Boston, USA
2 shared publications - Maria Gabrani
IBM Research Europe
2 shared publications - Adam Ivankay
IBM Research - Europe, Zurich, Switzerland
1 shared publication
Latest Publications
- Open Access
- 1 April 2026
- Published by Springer Nature
- p. 251-259
Decoding Self-paced Activity in Mice with ActionFormer
The locomotor activity (LMA) test is widely used to study spontaneous activity in rodents and the effects of drug treatments. However, standard automated metrics such as “distance traveled” often fail to capture the true complexity of behavior in this test, while manual annotations are labor intensive and prone to variability. Although recent advances in computational ethology, both unsupervised and supervised, have pushed forward analysis of spontaneous activity, challenges remain in providing interpretability and in temporal modeling. In this work, we adapt ActionFormer, a transformer-based architecture for temporal action localization, to analyze spontaneous activity of laboratory mice. ActionFormer significantly outperforms prior state-of-the-art methods in Average Precision (AP) across 7 common behaviors, and can capture characteristic drug-induced effects. By introducing behavior signatures, we showcase ActionFormer’s ability to uncover group differences and dose-response relationships in preclinical studies. Our findings demonstrate the potential of ActionFormer to further advance the frontier of computational ethology for laboratory rodents.
- 1 April 2026
- Published by Springer Nature
- p. 251-259
Open AccessDecoding Self-paced Activity in Mice with ActionFormer
The locomotor activity (LMA) test is widely used to study spontaneous activity in rodents and the effects of drug treatments. However, standard automated metrics such as “distance traveled” often fail to capture the true complexity of behavior in this test, while manual annotations are labor intensive and prone to variability. Although recent advances in computational ethology, both unsupervised and supervised, have pushed forward analysis of spontaneous activity, challenges remain in providing interpretability and in temporal modeling. In this work, we adapt ActionFormer, a transformer-based architecture for temporal action localization, to analyze spontaneous activity of laboratory mice. ActionFormer significantly outperforms prior state-of-the-art methods in Average Precision (AP) across 7 common behaviors, and can capture characteristic drug-induced effects. By introducing behavior signatures, we showcase ActionFormer’s ability to uncover group differences and dose-response relationships in preclinical studies. Our findings demonstrate the potential of ActionFormer to further advance the frontier of computational ethology for laboratory rodents.
- 21 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Imaging
- Vol. 43 (7) , 2599-2609
Multi-Scale Feature Alignment for Continual Learning of Unlabeled Domains
Methods for unsupervised domain adaptation (UDA) help to improve the performance of deep neural networks on unseen domains without any labeled data. Especially in medical disciplines such as histopathology, this is crucial since large datasets with detailed annotations are scarce. While the majority of existing UDA methods focus on the adaptation from a labeled source to a single unlabeled target domain, many real-world applications with a long life cycle involve more than one target domain. Thus, the ability to sequentially adapt to multiple target domains becomes essential. In settings where the data from previously seen domains cannot be stored, e.g., due to data protection regulations, the above becomes a challenging continual learning problem. To this end, we propose to use generative feature-driven image replay in conjunction with a dual-purpose discriminator that not only enables the generation of images with realistic features for replay, but also promotes feature alignment during domain adaptation. We evaluate our approach extensively on a sequence of three histopathological datasets for tissue-type classification, achieving state-of-the-art results. We present detailed ablation experiments studying our proposed method components and demonstrate a possible use-case of our continual UDA method for an unsupervised patch-based segmentation task given high-resolution tissue images. Our code is available at: https://github.com/histocartography/multi-scale-feature-alignment.
- 21 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Imaging
- Vol. 43 (7) , 2599-2609
Multi-Scale Feature Alignment for Continual Learning of Unlabeled Domains
Methods for unsupervised domain adaptation (UDA) help to improve the performance of deep neural networks on unseen domains without any labeled data. Especially in medical disciplines such as histopathology, this is crucial since large datasets with detailed annotations are scarce. While the majority of existing UDA methods focus on the adaptation from a labeled source to a single unlabeled target domain, many real-world applications with a long life cycle involve more than one target domain. Thus, the ability to sequentially adapt to multiple target domains becomes essential. In settings where the data from previously seen domains cannot be stored, e.g., due to data protection regulations, the above becomes a challenging continual learning problem. To this end, we propose to use generative feature-driven image replay in conjunction with a dual-purpose discriminator that not only enables the generation of images with realistic features for replay, but also promotes feature alignment during domain adaptation. We evaluate our approach extensively on a sequence of three histopathological datasets for tissue-type classification, achieving state-of-the-art results. We present detailed ablation experiments studying our proposed method components and demonstrate a possible use-case of our continual UDA method for an unsupervised patch-based segmentation task given high-resolution tissue images. Our code is available at: https://github.com/histocartography/multi-scale-feature-alignment.
- 8 August 2023
- Published by Elsevier in Medical Image Analysis
- Vol. 89, 102915
Weakly supervised joint whole-slide segmentation and classification in prostate cancer
- 8 August 2023
- Published by Elsevier in Medical Image Analysis
- Vol. 89, 102915
Weakly supervised joint whole-slide segmentation and classification in prostate cancer
- Open Access
- 7 August 2023
- Published by Elsevier in Medical Image Analysis
- Vol. 89, 102924
Generative appearance replay for continual unsupervised domain adaptation
- 7 August 2023
- Published by Elsevier in Medical Image Analysis
- Vol. 89, 102924
Open AccessGenerative appearance replay for continual unsupervised domain adaptation
- 23 October 2022
- Published by Springer Nature
- p. 699-715
Differentiable Zooming for Multiple Instance Learning on Whole-Slide Images
- 23 October 2022
- Published by Springer Nature
- p. 699-715
Differentiable Zooming for Multiple Instance Learning on Whole-Slide Images
- 1 March 2022
- Published by Elsevier in Smart Health
- Vol. 23
Privacy preserving synthetic respiratory sounds for class incremental learning
- 1 March 2022
- Published by Elsevier in Smart Health
- Vol. 23
Privacy preserving synthetic respiratory sounds for class incremental learning
- Open Access
Artificial Intelligence Decision Support for Medical Triage
Applying state-of-the-art machine learning and natural language processing on approximately one million of teleconsultation records, we developed a triage system, now certified and in use at the largest European telemedicine provider. The system evaluates care alternatives through interactions with patients via a mobile application. Reasoning on an initial set of provided symptoms, the triage application generates AI-powered, personalized questions to better characterize the problem and recommends the most appropriate point of care and time frame for a consultation. The underlying technology was developed to meet the needs for performance, transparency, user acceptance and ease of use, central aspects to the adoption of AI-based decision support systems. Providing such remote guidance at the beginning of the chain of care has significant potential for improving cost efficiency, patient experience and outcomes. Being remote, always available and highly scalable, this service is fundamental in high demand situations, such as the current COVID-19 outbreak.
Open AccessArtificial Intelligence Decision Support for Medical Triage
Applying state-of-the-art machine learning and natural language processing on approximately one million of teleconsultation records, we developed a triage system, now certified and in use at the largest European telemedicine provider. The system evaluates care alternatives through interactions with patients via a mobile application. Reasoning on an initial set of provided symptoms, the triage application generates AI-powered, personalized questions to better characterize the problem and recommends the most appropriate point of care and time frame for a consultation. The underlying technology was developed to meet the needs for performance, transparency, user acceptance and ease of use, central aspects to the adoption of AI-based decision support systems. Providing such remote guidance at the beginning of the chain of care has significant potential for improving cost efficiency, patient experience and outcomes. Being remote, always available and highly scalable, this service is fundamental in high demand situations, such as the current COVID-19 outbreak.
- 2 October 2018
- Published by Springer Nature
- p. 312-319
A Structure-Aware Convolutional Neural Network for Skin Lesion Classification
- 2 October 2018
- Published by Springer Nature
- p. 312-319
A Structure-Aware Convolutional Neural Network for Skin Lesion Classification