Ibrahim Ethem Hamamci
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Ibrahim Ethem Hamamci has a total of 256 co-authors. Here are the top co-authors by number of shared publications:
Sezgin Er
Istanbul Medipol University
7 shared publicationsAnjany Sekuboyina
Department of Quantitative Biomedicine, University of Zurich, Zurich, Switzerland
6 shared publications- Bjoern Menze
Department of Quantitative Biomedicine, University of Zurich, Zurich, Switzerland
6 shared publications Bjoern Menze
University of Zürich
5 shared publicationsEnis Simsar
Department of Computer Science, ETH Zurich, Zurich, Switzerland
5 shared publications- Bastian Wittmann
Department of Quantitative Biomedicine, University of Zurich, Zurich, Switzerland
4 shared publications
Latest Publications
- Open Access
- 28 October 2026
- Published by Springer Nature in Nature Communications
CADS: a comprehensive anatomical dataset and segmentation for whole-body anatomy in computed tomography
Volumetric computed tomography (CT) segmentation supports quantitative image analysis, but most AI-based methods remain organ-specific, with fragmented models, inconsistent label definitions, and heterogeneous training and evaluation protocols. Here we present CADS, a unified data–model framework for scalable and generalizable whole-body CT segmentation. We construct a SNOMED-CT-referenced anatomical label space harmonizing heterogeneous cohorts from over 100 imaging centers, yielding complete annotations of 167 structures across 22,022 CT volumes and markedly expanding anatomical coverage and scale beyond existing resources. To complete labels on sparsely annotated CTs, we combine self-training with shape-informed quality control based on neural implicit functions, curating high-quality pseudo-labels. Trained on this resource, CADS shows robust whole-body segmentation across distribution shifts and improved performance relative to state-of-the-art approaches across 18 public datasets. We further validate it in an independent, expert-reviewed real-world oncology cohort of 2864 CTs. The curated dataset, pretrained models, and user-friendly software together provide an accessible, reproducible foundation for large-scale anatomical segmentation and computational anatomy research.
- 28 October 2026
- Published by Springer Nature in Nature Communications
Open AccessCADS: a comprehensive anatomical dataset and segmentation for whole-body anatomy in computed tomography
Volumetric computed tomography (CT) segmentation supports quantitative image analysis, but most AI-based methods remain organ-specific, with fragmented models, inconsistent label definitions, and heterogeneous training and evaluation protocols. Here we present CADS, a unified data–model framework for scalable and generalizable whole-body CT segmentation. We construct a SNOMED-CT-referenced anatomical label space harmonizing heterogeneous cohorts from over 100 imaging centers, yielding complete annotations of 167 structures across 22,022 CT volumes and markedly expanding anatomical coverage and scale beyond existing resources. To complete labels on sparsely annotated CTs, we combine self-training with shape-informed quality control based on neural implicit functions, curating high-quality pseudo-labels. Trained on this resource, CADS shows robust whole-body segmentation across distribution shifts and improved performance relative to state-of-the-art approaches across 18 public datasets. We further validate it in an independent, expert-reviewed real-world oncology cohort of 2864 CTs. The curated dataset, pretrained models, and user-friendly software together provide an accessible, reproducible foundation for large-scale anatomical segmentation and computational anatomy research.
- 25 September 2026
- Published by Springer Nature
- p. 122-132
CTTok: Voxel-Abulary for Autoregressive 3D CT Volume Generation with Large Language Models
- 25 September 2026
- Published by Springer Nature
- p. 122-132
CTTok: Voxel-Abulary for Autoregressive 3D CT Volume Generation with Large Language Models
- 17 September 2026
- Published by Springer Nature
- p. 388-406
Comprehensive Language–Image Pre-training for 3D Medical Image Understanding
- 17 September 2026
- Published by Springer Nature
- p. 388-406
Comprehensive Language–Image Pre-training for 3D Medical Image Understanding
- 23 July 2026
- Published by Massachusetts Medical Society in NEJM AI
- Vol. 3 (8)
The TopCoW Challenge — Topology-Aware Circle of Willis Segmentation for CT and MR Angiography
- 23 July 2026
- Published by Massachusetts Medical Society in NEJM AI
- Vol. 3 (8)
The TopCoW Challenge — Topology-Aware Circle of Willis Segmentation for CT and MR Angiography
- Open Access
- 15 July 2026
- Published by Springer Nature in Research Square
Large-Scale Multi-Cancer Detection by Learning Segmentation from Reports
More than 300 million computed tomography (CT) scans are performed worldwide each year, yet many early or incidental tumors in these scans remain undetected. Artificial intelligence (AI) could help: segmentation models can surpass radiologists and alternative AI models in detecting tumors, and they localize the tumors for radiologist verification. However, segmentation-based tumor detection has long been limited by the need for tumor masks: radiologist-drawn tumor outlines that are scarce, expensive, and entirely unavailable for many cancer types. In contrast, nearly every CT scan is accompanied by a radiology report with detailed tumor descriptions. Yet, these reports have not been used effectively to train tumor segmentation models. Here, we introduce R-Super, a framework that converts routine radiology and pathology reports into localized training signals for tumor segmentation. R-Super trains AI to segment tumors that match their descriptions in reports. Reports are only needed for training, not inference. We trained R-Super on 127,496 CT-Report pairs (42 million 2D images, USA) and evaluated it internally at UCSF (N=2,301, USA) and externally at Stanford (N=1,976, USA), Medipol (N=1,327, Turkey), and Basel (N=2,935, Switzerland). R-Super detects 7 tumor types for which public tumor masks are scarce or absent: spleen, gallbladder, prostate, bladder, uterus, esophagus, and adrenal tumors. Training R-Super on over 100,000 reports (no mask) outperformed mask-based segmentation models trained on 870 masks, demonstrating that large-scale report-based training surpasses smaller-scale mask-based training. Alternatively, by training R-Super on both these reports and masks together, cancer detection sensitivity increased by over +11% beyond mask-only training, and DSC by +14%. R-Super significantly surpassed 6 alternative training frameworks trained on the same dataset and 9 leading public AI models. R-Super significantly surpassed six radiologists in detecting six tumor types and matched them for uterus tumors in a reader study. On average, R-Super detected 56% more malignant tumors than the radiologists, at matched false positive rate. These results show that radiology reports are not merely clinical documentation, but a large-scale, underutilized source of localized supervision for training more accurate cancer detection AI. By effectively learning from reports, R-Super enables tumor segmentation models to scale beyond scarce radiologist-drawn tumor masks and advances automated and incidental cancer detection closer to clinical deployment. We release code, over 22,000 CT scans and reports, and the first public AI model to reach or surpass radiologist performance in detecting these tumor types on CT.
- 15 July 2026
- Published by Springer Nature in Research Square
Open AccessLarge-Scale Multi-Cancer Detection by Learning Segmentation from Reports
More than 300 million computed tomography (CT) scans are performed worldwide each year, yet many early or incidental tumors in these scans remain undetected. Artificial intelligence (AI) could help: segmentation models can surpass radiologists and alternative AI models in detecting tumors, and they localize the tumors for radiologist verification. However, segmentation-based tumor detection has long been limited by the need for tumor masks: radiologist-drawn tumor outlines that are scarce, expensive, and entirely unavailable for many cancer types. In contrast, nearly every CT scan is accompanied by a radiology report with detailed tumor descriptions. Yet, these reports have not been used effectively to train tumor segmentation models. Here, we introduce R-Super, a framework that converts routine radiology and pathology reports into localized training signals for tumor segmentation. R-Super trains AI to segment tumors that match their descriptions in reports. Reports are only needed for training, not inference. We trained R-Super on 127,496 CT-Report pairs (42 million 2D images, USA) and evaluated it internally at UCSF (N=2,301, USA) and externally at Stanford (N=1,976, USA), Medipol (N=1,327, Turkey), and Basel (N=2,935, Switzerland). R-Super detects 7 tumor types for which public tumor masks are scarce or absent: spleen, gallbladder, prostate, bladder, uterus, esophagus, and adrenal tumors. Training R-Super on over 100,000 reports (no mask) outperformed mask-based segmentation models trained on 870 masks, demonstrating that large-scale report-based training surpasses smaller-scale mask-based training. Alternatively, by training R-Super on both these reports and masks together, cancer detection sensitivity increased by over +11% beyond mask-only training, and DSC by +14%. R-Super significantly surpassed 6 alternative training frameworks trained on the same dataset and 9 leading public AI models. R-Super significantly surpassed six radiologists in detecting six tumor types and matched them for uterus tumors in a reader study. On average, R-Super detected 56% more malignant tumors than the radiologists, at matched false positive rate. These results show that radiology reports are not merely clinical documentation, but a large-scale, underutilized source of localized supervision for training more accurate cancer detection AI. By effectively learning from reports, R-Super enables tumor segmentation models to scale beyond scarce radiologist-drawn tumor masks and advances automated and incidental cancer detection closer to clinical deployment. We release code, over 22,000 CT scans and reports, and the first public AI model to reach or surpass radiologist performance in detecting these tumor types on CT.
- 25 June 2026
- Published by Massachusetts Medical Society in NEJM AI
- Vol. 3 (7)
ReXGroundingCT: A 3D Chest CT Dataset for Segmentation of Findings from Free-Text Reports
- 25 June 2026
- Published by Massachusetts Medical Society in NEJM AI
- Vol. 3 (7)
ReXGroundingCT: A 3D Chest CT Dataset for Segmentation of Findings from Free-Text Reports
- Top Cited PapersOpen Access
- 12 February 2026
- Published by Springer Nature in Nature Biomedical Engineering
- Vol. 10 (8) , 1610-1628
Generalist foundation models from a multimodal dataset for 3D computed tomography
Advancements in medical imaging AI, particularly in 3D imaging, have been limited due to the scarcity of comprehensive datasets. We introduce CT-RATE, a public dataset that pairs 3D medical images with corresponding textual reports. CT-RATE comprises 25,692 non-contrast 3D chest CT scans from 21,304 unique patients. Each scan is accompanied by its corresponding radiology report. Leveraging CT-RATE, we develop CT-CLIP, a CT-focused contrastive language–image pretraining framework designed for broad applications without the need for task-specific training. We demonstrate how CT-CLIP can be used in multi-abnormality detection and case retrieval, and outperforms state-of-the-art fully supervised models across all key metrics. By combining CT-CLIP’s vision encoder with a pretrained large language model, we create CT-CHAT, a vision–language foundational chat model for 3D chest CT volumes. Fine-tuned on over 2.7 million question–answer pairs derived from the CT-RATE dataset, CT-CHAT underscores the necessity for specialized methods in 3D medical imaging. Collectively, the open-source release of CT-RATE, CT-CLIP and CT-CHAT not only addresses critical challenges in 3D medical imaging but also lays the groundwork for future innovations in medical AI and improved patient care.
- 12 February 2026
- Published by Springer Nature in Nature Biomedical Engineering
- Vol. 10 (8) , 1610-1628
Top Cited PapersOpen AccessGeneralist foundation models from a multimodal dataset for 3D computed tomography
Advancements in medical imaging AI, particularly in 3D imaging, have been limited due to the scarcity of comprehensive datasets. We introduce CT-RATE, a public dataset that pairs 3D medical images with corresponding textual reports. CT-RATE comprises 25,692 non-contrast 3D chest CT scans from 21,304 unique patients. Each scan is accompanied by its corresponding radiology report. Leveraging CT-RATE, we develop CT-CLIP, a CT-focused contrastive language–image pretraining framework designed for broad applications without the need for task-specific training. We demonstrate how CT-CLIP can be used in multi-abnormality detection and case retrieval, and outperforms state-of-the-art fully supervised models across all key metrics. By combining CT-CLIP’s vision encoder with a pretrained large language model, we create CT-CHAT, a vision–language foundational chat model for 3D chest CT volumes. Fine-tuned on over 2.7 million question–answer pairs derived from the CT-RATE dataset, CT-CHAT underscores the necessity for specialized methods in 3D medical imaging. Collectively, the open-source release of CT-RATE, CT-CLIP and CT-CHAT not only addresses critical challenges in 3D medical imaging but also lays the groundwork for future innovations in medical AI and improved patient care.
- Open Access
Better Tokens for Better 3D: Advancing Vision-Language Modeling in 3D Medical Imaging
Recent progress in vision-language modeling for 3D medical imaging has been fueled by large-scale computed tomography (CT) corpora with paired free-text reports, stronger architectures, and powerful pretrained models. This has enabled applications such as automated report generation and text-conditioned 3D image synthesis. Yet, current approaches struggle with high-resolution, long-sequence volumes: contrastive pretraining often yields vision encoders that are misaligned with clinical language, and slice-wise tokenization blurs fine anatomy, reducing diagnostic performance on downstream tasks. We introduce BTB3D (Better Tokens for Better 3D), a causal convolutional encoder-decoder that unifies 2D and 3D training and inference while producing compact, frequency-aware volumetric tokens. A three-stage training curriculum enables (i) local reconstruction, (ii) overlapping-window tiling, and (iii) long-context decoder refinement, during which the model learns from short slice excerpts yet generalizes to scans exceeding 300 slices without additional memory overhead. BTB3D sets a new state-of-the-art on two key tasks: it improves BLEU scores and increases clinical F1 by 40% over CT2Rep, CT-CHAT, and Merlin for report generation; and it reduces FID by 75% and halves FVD compared to GenerateCT and MedSyn for text-to-CT synthesis, producing anatomically consistent 512*512*241 volumes. These results confirm that precise three-dimensional tokenization, rather than larger language backbones alone, is essential for scalable vision-language modeling in 3D medical imaging. The codebase is available at: https://github.com/ibrahimethemhamamci/BTB3D
Open AccessBetter Tokens for Better 3D: Advancing Vision-Language Modeling in 3D Medical Imaging
Recent progress in vision-language modeling for 3D medical imaging has been fueled by large-scale computed tomography (CT) corpora with paired free-text reports, stronger architectures, and powerful pretrained models. This has enabled applications such as automated report generation and text-conditioned 3D image synthesis. Yet, current approaches struggle with high-resolution, long-sequence volumes: contrastive pretraining often yields vision encoders that are misaligned with clinical language, and slice-wise tokenization blurs fine anatomy, reducing diagnostic performance on downstream tasks. We introduce BTB3D (Better Tokens for Better 3D), a causal convolutional encoder-decoder that unifies 2D and 3D training and inference while producing compact, frequency-aware volumetric tokens. A three-stage training curriculum enables (i) local reconstruction, (ii) overlapping-window tiling, and (iii) long-context decoder refinement, during which the model learns from short slice excerpts yet generalizes to scans exceeding 300 slices without additional memory overhead. BTB3D sets a new state-of-the-art on two key tasks: it improves BLEU scores and increases clinical F1 by 40% over CT2Rep, CT-CHAT, and Merlin for report generation; and it reduces FID by 75% and halves FVD compared to GenerateCT and MedSyn for text-to-CT synthesis, producing anatomically consistent 512*512*241 volumes. These results confirm that precise three-dimensional tokenization, rather than larger language backbones alone, is essential for scalable vision-language modeling in 3D medical imaging. The codebase is available at: https://github.com/ibrahimethemhamamci/BTB3D
- 2 November 2024
- Published by Springer Nature
- p. 126-143
GenerateCT: Text-Conditional Generation of 3D Chest CT Volumes
- 2 November 2024
- Published by Springer Nature
- p. 126-143
GenerateCT: Text-Conditional Generation of 3D Chest CT Volumes
- Open Access
- 28 October 2024
- Published by Springer Nature in Research Square
Developing Generalist Foundation Models from a Multimodal Dataset for 3D Computed Tomography
While computer vision has achieved tremendous success with multimodal encoding and direct textual interaction with images via chat-based large language models, similar advancements in medical imaging AI—particularly in 3D imaging—have been limited due to the scarcity of comprehensive datasets. To address this critical gap, we introduce CT-RATE, the first dataset that pairs 3D medical images with corresponding textual reports. CT-RATE comprises 25,692 non-contrast 3D chest CT scans from 21,304 unique patients. Through various reconstructions, these scans are expanded to 50,188 volumes, totaling over 14.3 million 2D slices. Each scan is accompanied by its corresponding radiology report. Leveraging CT-RATE, we develop CT-CLIP, a CT-focused contrastive language-image pretraining framework designed for broad applications without the need for task-specific training. We demonstrate how CT-CLIP can be used in two tasks: multi-abnormality detection and case retrieval. Remarkably, in multi-abnormality detection, CT-CLIP outperforms state-of-the-art fully supervised models across all key metrics, effectively eliminating the need for manual annotation. In case retrieval, it efficiently retrieves relevant cases using either image or textual queries, thereby enhancing knowledge dissemination. By combining CT-CLIP's vision encoder with a pretrained large language model, we create CT-CHAT, a vision-language foundational chat model for 3D chest CT volumes. Finetuned on over 2.7 million question-answer pairs derived from the CT-RATE dataset, CT-CHAT surpasses other multimodal AI assistants, underscoring the necessity for specialized methods in 3D medical imaging. Collectively, the open-source release of CT-RATE, CT-CLIP, and CT-CHAT not only addresses critical challenges in 3D medical imaging but also lays the groundwork for future innovations in medical AI and improved patient care.
- 28 October 2024
- Published by Springer Nature in Research Square
Open AccessDeveloping Generalist Foundation Models from a Multimodal Dataset for 3D Computed Tomography
While computer vision has achieved tremendous success with multimodal encoding and direct textual interaction with images via chat-based large language models, similar advancements in medical imaging AI—particularly in 3D imaging—have been limited due to the scarcity of comprehensive datasets. To address this critical gap, we introduce CT-RATE, the first dataset that pairs 3D medical images with corresponding textual reports. CT-RATE comprises 25,692 non-contrast 3D chest CT scans from 21,304 unique patients. Through various reconstructions, these scans are expanded to 50,188 volumes, totaling over 14.3 million 2D slices. Each scan is accompanied by its corresponding radiology report. Leveraging CT-RATE, we develop CT-CLIP, a CT-focused contrastive language-image pretraining framework designed for broad applications without the need for task-specific training. We demonstrate how CT-CLIP can be used in two tasks: multi-abnormality detection and case retrieval. Remarkably, in multi-abnormality detection, CT-CLIP outperforms state-of-the-art fully supervised models across all key metrics, effectively eliminating the need for manual annotation. In case retrieval, it efficiently retrieves relevant cases using either image or textual queries, thereby enhancing knowledge dissemination. By combining CT-CLIP's vision encoder with a pretrained large language model, we create CT-CHAT, a vision-language foundational chat model for 3D chest CT volumes. Finetuned on over 2.7 million question-answer pairs derived from the CT-RATE dataset, CT-CHAT surpasses other multimodal AI assistants, underscoring the necessity for specialized methods in 3D medical imaging. Collectively, the open-source release of CT-RATE, CT-CLIP, and CT-CHAT not only addresses critical challenges in 3D medical imaging but also lays the groundwork for future innovations in medical AI and improved patient care.