Featured
-
-
Brief Communication
| Open Access
A reinforcement learning model for AI-based decision support in skin cancer
A reinforcement learning model developed to adapt artificial intelligence (AI) predictions to human preferences showed better sensitivity for skin cancer diagnoses and improved management decisions compared to a supervised learning model.
- Catarina Barata
- , Veronica Rotemberg
- & Harald Kittler
-
Brief Communication |
Artificial-intelligence-based molecular classification of diffuse gliomas using rapid, label-free optical imaging
DeepGlioma, a multimodal deep learning approach for intraoperative diagnostic screening of diffuse glioma, trained on stimulated Raman histology and large-scale public genomic data, can predict molecular alterations for diffuse glioma diagnosis with high accuracy.
- Todd Hollon
- , Cheng Jiang
- & Daniel A. Orringer
-
Article
| Open Access
Whole-body CD8+ T cell visualization before and during cancer immunotherapy: a phase 1/2 trial
A CD8-specific, one-armed antibody positron emission tomography tracer enables the visualization of the immune response in patients with solid tumors before and after starting immunotherapy.
- Laura Kist de Ruijter
- , Pim P. van de Donk
- & Elisabeth G. E. de Vries
-
Correspondence |
Bridging the gap with the UK Genomics Pathology Imaging Collection
- Charlotte N. Jennings
- , Matthew P. Humphries
- & Darren Treanor
-
Article
| Open Access
Swarm learning for decentralized artificial intelligence in cancer histopathology
A decentralized, privacy-preserving machine learning framework used to train a clinically relevant AI system identifies actionable molecular alterations in patients with colorectal cancer by use of routine histopathology slides collected in real-world settings.
- Oliver Lester Saldanha
- , Philip Quirke
- & Jakob Nikolas Kather
-
Letter |
Near real-time intraoperative brain tumor diagnosis using stimulated Raman histology and deep neural networks
A prospective, multicenter, case–control clinical trial evaluates the potential of artificial intelligence for providing accurate bedside diagnosis of patients with brain tumors.
- Todd C. Hollon
- , Balaji Pandian
- & Daniel A. Orringer
-
Letter |
Deep learning-based classification of mesothelioma improves prediction of patient outcome
Deep convolutional neural networks predict survival of mesothelioma patients and identify histological features associated with outcome that transcend current histological classifications.
- Pierre Courtiol
- , Charles Maussion
- & Thomas Clozel
-
Brief Communication |
Deep learning can predict microsatellite instability directly from histology in gastrointestinal cancer
A deep residual learning framework identifies microsatellite instability in histology slides from patients with cancer and can be used to guide immunotherapy.
- Jakob Nikolas Kather
- , Alexander T. Pearson
- & Tom Luedde
-
Letter |
89Zr-atezolizumab imaging as a non-invasive approach to assess clinical response to PD-L1 blockade in cancer
Initial results from a first-in-human study show that PET imaging with PD-L1 antibodies outperforms immunohistochemistry- or RNA-sequencing-based biomarkers for prediction of clinical response to immunotherapy.
- Frederike Bensch
- , Elly L. van der Veen
- & Elisabeth G. E. de Vries
-
Technical Report |
Microscopic lymph node tumor burden quantified by macroscopic dual-tracer molecular imaging
Tichauer et al. describe a dual-tracer approach to quantify cancer cell receptor concentrations, in this case epidermal growth factor receptor, in lymph nodes, that can also correct for nonspecific uptake.
- Kenneth M Tichauer
- , Kimberley S Samkoe
- & Brian W Pogue
-
Technical Report |
In vivo proteomic imaging analysis of caveolae reveals pumping system to penetrate solid tumors
Proteomic-imaging analysis of caveolae shows active transvascular pumping of antibodies across the endothelial cell barrier and into solid tumors against a concentration gradient.
- Phil Oh
- , Jacqueline E Testa
- & Jan E Schnitzer
-
Technical Report |
Magnetic resonance imaging of tumor glycolysis using hyperpolarized 13C-labeled glucose
One of the most likely substrates for metabolic imaging of response to treatment in cancer is glucose, but until now, using hyperpolarized 13C-labelled glucose has been problematic because of the short lifetime of the hyperpolarization in this molecule. Using [U-13C, U-2H]glucose, Tiago Rodrigues et al. now show that they are able to image its glycolytic conversion to lactate in two mouse tumor models in vivo, and that in one model, flux is markedly reduced after treatment with the chemotherapeutic drug etoposide.
- Tiago B Rodrigues
- , Eva M Serrao
- & Kevin M Brindle
-
Technical Report |
Tumor-specific imaging through progression elevated gene-3 promoter-driven gene expression
Bhang and colleagues have developed a tumor-specific imaging strategy that uses the progression elevated gene-3 (PEG-3) promoter, known to be specifically associated with malignant transformation, to selectively drive the expression of luciferase or herpes simplex virus 1 thymidine kinase reporters. Systemic delivery of PEG-3 promoter–driven constructs using a nonviral gene delivery vehicle allowed detection of both primary tumors and micrometastatic disease in mouse models of human melanoma and breast cancer.
- Hyo-eun C Bhang
- , Kathleen L Gabrielson
- & Martin G Pomper
